Smarter DevOps - Why You Need Continuous Monitoring
By Yann Guernion

Those from a dev-centric background might discount monitoring from their DevOps approach. But if you're already automating deployment and testing, why wouldn't you use the right set of tools to avoid manual performance monitoring?

Of course, you do performance tests each time you release changes to pre-production. You're checking for response time, processor utilization, memory occupation, I/O activity and so on, but how can you ensure normal behavior without a reference point? In other words - despite your good intentions - aren't you using critical production as a first level of benchmark?

If you run pre-production performance tests regularly and you can compare those results to what is seen in production, there is a greater chance of detecting the processor utilization jump that caused slowdowns and then relate it to a code change. But the reality is that most developers lack the ability to monitor how their code performs in the test, staging and production environments.

DevOps teams then face significant challenges in guaranteeing expected application behavior:

• Understanding application performance before and after new code is pushed and pinpointing defects early, before they spread.
• Diving back into the history of deployments, determining the impact on the infrastructure throughput and response time.
• Forecasting infrastructure utilization bottlenecks due to changes into the code or variations in the workload.

These difficulties can be avoided with continuous monitoring that allows developers to see performance in production without exposing critical assets. Continuous monitoring tools provide highly granular real-time and historical performance data without putting extra load on the production environment. Rather than waiting for production performance data to analyze what went wrong, the DevOps team is able to develop performance analytics models that can anticipate operational and quality problems before the delivery phase.

When you adopt a continuous monitoring approach and it's associated tools, Dev and Ops then have a basis for sharing and comparing results, evaluating the progress and tracking the real impact of changes made.  With software developers and systems engineers having a common basis for discussion and progress evaluation, everyone is in better shape to work towards their common goal - practicing smarter DevOps.